The Performance of Homeric Epic

Thunder but no Clouds: the Genesis of the Homeric Text

Richard Janko
University College, London
E-mail: R.Janko@ucl.ac.uk

'The more I understand the South slavic poetry and the nature of the unity of the oral poem, the clearer it seems to me that the Iliad and the Odyssey are very exactly, as we have them, each one ot them the rounded and finished work of a single singer . . . I even figure to myself, just now, the moment when the author of the Odyssey sat and dictated his song, while another, with writing materials, wrote it down verse by verse, even in the way that our singers sit in the immobility of their thought, watching the motion of Nikola's hand across the empty page, when it will tell them it is the instant for them to speak the next verse.' So wrote Milman Parry late in his life.(1) For this particular commentator on Homer's Iliad, Parry's hypothesis about the origins of the Homeric text influenced almost every line I wrote. Writing the Commentary only strengthened my view that the Iliad and Odyssey are texts orally composed in performance, written down by dictation from that same performance. What kind of text we are dealing with matters far more, for editing and interpreting Homer, than other questions, like whether the Iliad and Odyssey are the creations of one single poet (they are, in my view), or when the poems were created.

To digress for a moment on their date, the linguistic evidence offered by G.P. Edwards (2) and myself that the texts of these epics were fixed before those of Hesiod's poems is fundamental; this method disproved the claim that the choice of genre has a significant effect on the stage of linguistic evolution observable in any given text of the early epic tradition. I remain unconvinced by attempts to date Homer to the seventh or sixth centuries. (3) On the contrary, the recent arguments of C.J. Ruijgh, (4) and the existence of the new alphabetic inscription from Gabii dated to c. 800 B.C., incline me to place the epics even earlier than I used to, to c. 775-750 B.C. for the Iliad and slightly later for the Odyssey.

In the last decade a wide variety of conflicting hypotheses have continued to be advanced about the genesis of the text of Homer, notably pen-and-paper creation by the poet or poets, and a process of textual fixation, centred on Pisistratean Athens, of traditions of oral performance diffused throughout Greece. This paper will argue that evidence internal to the text itself shows that the Homeric poems are direct products of oral composition in performance, and were written down by a contemporary amanuensis from dictation by the poet. In conclusion I shall offer some ideas about how this recording took place.

The researches of Parry and Lord, including Lord's two new books, have demonstrated that the great epics of Homer are most unlikely to have been composed by a poet using writing in any conventional way. Their work explained at a stroke a whole set of disparate phenomena, not only the extraordinary existence of Homer's extensive and economical systems of set formulae, but also its frequent metrical irregularities, the epic diction derived from several dialects, the near-repetition of standard scene-patterns, and minor inconsistencies of content. These features are typical of oral narrative epics, and many of them help the singers sustain their flow of composition, as Parry and Lord showed from their field-work in Bosnia. Such songs are neither improvised outright nor memorized: the singers use their memory and knowledge of traditional materials and patterns to retell in each new singing a traditional tale. The work of the best bards, like Avdo Mededovic, is far more creative within the tradition than that of the inferior singers whose work Lord also published. There was much ancient Greek oral poetry of equally poor quality, such as the Shield of Heracles.

Many Homerists have been unaware that, in criticising the oralist approach, they have been aiming at a moving target. One can now cite Lord's own words to show how some aspects of the work of Parry and Lord have been modified since The Singer of Tales, and others have been misunderstood.

First, in The Singer of Tales the impression was given that oral composition is the same as oral epic composition. Lord denies this explicitly in his recent books, in which he studies ballads and lyrics from many cultures. However, his sweeping use of the term allowed R. Finnegan to claim (5) that anything orally disseminated is an oral text, a claim that is self-evidently true, but only in a trivial sense. Yet this claim has widely been seen as an answer to their doubts, by those unable or unwilling to grasp how illiterate singers can compose and perform coherent narratives thousands of lines in length. Oral performance is not to be confused with oral composition-in-performance.

Also, Lord is often understood as having said that an oral tradition is killed off at once by the introduction of writing. He has recently offered a much more careful formulation: 'literacy carries the seeds of the eventual demise of oral traditional composition . . . It is not, however, writing per se that brings about the change; traditional oral epic flourished in the Slavic Balkans for centuries in communities where significant portions of the population were literate. But gradually the epic came to be written down, and the concept of a fixed text, and of the text, of a song came to be current. With that concept arose the need for memorization rather than recomposition as a means of transmission'. Again, everyone has used the term 'Parry-Lord theory' or 'oral theory'. Lord has now protested against this:

the phrase 'oral theory' with regard to the investigations into South Slavic oral epic by Parry and me is a misnomer. These findings do not constitute a 'theory'; rather, they provide demonstrated facts concerning oral traditional poetry. . . . Where else but to a tradition continuing into modern times could a scholar go to look for clues to the nature of epics such as the Homeric poems and Beowulf, the method of whose composition is not documented and is subject to controversy?

Again, the basis of oral composition is not improvisation ex nihilo-- quite the contrary. To be able to improvise, the poet must first master the oral-traditional poetic language, with enough of the traditional scenes and story-patterns to enable him to hold his tale together. Nor does oral composition preclude careful rehearsal. (6) Lord records (7) that a Bosnian singer claimed to practise his songs while tending his sheep. Indeed, for a poet composing as he sings, every performance is a rehearsal for the next time. Songs can have a relatively stable existence in the mind of a particular singer: thus one of them sang the 'same song' twice, with an interval of seventeen years between the two performances, but with striking similarities between them, even to the extent of the singer making the same mistake in plot-construction. (8)

Again, it is mistaken to hold that memorization plays any part in the composition of oral epic poetry. (9) Oral literature is only memorized when there exists an idea of a fixed text; once the texts of the Homeric and Hesiodic poems had been fixed in writing, the way was open to memorize them, which is what the rhapsodes did. But the idea of a fixed text could not exist without writing, at least until the invention of the phonograph. Oral poets remember how to compose oral poetry, which is quite a different thing. (10) They remember a huge stock of traditional expressions, verses, blocks of verses, typical scenes and story-patterns. A good bard will have an excellent memory, but he does not remember the text of what he performs. Formulae are not, as has been thought, an aid to memorization: quite the reverse. Formulae exist as an aid to composition, not to memorization. (11) They are memorable phrases, but they are not memorized in the same way as we memorize telephone numbers.

The last and worst misconception is that oral literature of high quality is an impossibility. The paradox of such great literature--for we cannot call the Homeric poems anything else--in an oral formular style has inevitably provoked the compromise, accepted by many, that, although Homer came at the end of a long oral tradition, his poems are so good that they must have been composed with the aid of writing; they are 'oral- derived'. In fact this is not a reasonable and moderate compromise, but rests on an unexamined assumption, commonly made by literate and illiterate people alike, that written literature must be superior to oral literature. This is false: there is good and bad oral literature, just as there is good and bad written literature. Oral and written literature differ in certain ways, but we cannot assume a priori that the oral poets of the past were incapable of managing large structures like the story-patterns of the Iliad and Odyssey, nor that they were incapable of achieving powerful effects in detail.

Even if one believes, as I do, that the Iliad and Odyssey must be oral dictated texts, there is much to be said in favour of premeditation and writing as aids to Homeric composition. Lord showed that once a poet is accustomed to the slower pace of dictation, he can take advantage of it to improve the quality of his performance, not only in terms of avoiding defective lines, but also in more careful premeditation of the story itself. (12) But poets composing orally cannot go back and alter what they have composed. If the Homeric poems were in fact composed with the help of writing, we would expect them to exhibit a much smoother surface and a much more self-conscious style than they have. But can we move from probable hypothesis to observing the process of dictation at work in the texts? I believe that we can, frequently.

We can be certain that Homer did not use writing to plan his texts. He included frequent summaries and predictions of events partly to help him solidify and recall the plan of his story, and his numerous catalogues of warriors served in part to establish in his mind who was to be killed off in the next few hundred verses. Nor did he use it to correct his texts. Vestiges of this failure to use writing, which I believe only make sense as vestiges of dictation, can be discerned wherever one looks. At Iliad XIV 170ff. Hera is adorning herself:

ambrosiei men proton apo khoros himeroentos 170
lymata panta katheren, aleipsato de lip' elaioi 171
ambrosioi hedanoi, to rha hoi tethumenon een. 172
tou kai kinumenoio Dios kata khalkobates do 173
empes es gaian te kai ouranon hiket' autme. 174
toi rh' he ge khroa kalon aleipsamene ide khaitas 175
peksamene khersi plokamous eplekse phaeinous 176
kalous ambrosious ek kraatos athanatoio. 177
amphi d' ar' ambrosion heanon hesasth', hon hoi Athene

Now the repetition of ambrosios four times in lines 170-178 is certainly not a feature favoured in 'literary' poetry, and the third instance of the word, kalous ambriosous in line 177, was emended away by Zenodotus and Aristophanes, who read kalous kai megathous, although it was left alone by Aristarchus. Not one manuscript, by the way, has the reading of the earlier Alexandrians. A different problem arises in lines 173-4:

tou kai kinumenoio Dios poti khalkobates do
empes es gaian te kai ouranon hiket' autme

Aristarchus altered poti into kata, which makes better sense; how can the poet say 'when Hera's dress was moved to Zeus' house, its perfume reached heaven and earth'? Aristarchus is quite right to want the text to say 'when Hera's dress was moved in Zeus' house, its perfume reached heaven and earth'. But that is not what the text says; the poet has anticipated the idea of the perfume reaching heaven (the same place as Zeus' house) and uttered the wrong preposition, since his thoughts again ran ahead of his tongue. Aristarchus, like any good editor of a written text, has corrected the error. But a scrupulous editor of an orally dictated text--which Aristarchus did not know that he had before him-- would not alter this, but would remark it in a note. As usual, the orally dictated imperfection is still faithfully preserved by the manuscripts; in terms of textual criticism, it has the status of a lectio difficilior.

The Alexandrians' procedure is paralleled in the Parry Collection. During one of my visits to it Lord showed me transcripts of Bosnian songs edited for publication by the collector Marjanovic in the late nineteenth century. I quote his description of them:

Marjanovic made many changes in the manuscript. He left out lines and added lines; he left out blocks of five to ten lines. He changed all eleven-syllable lines to ten syllables, and sometimes he combined two lines. His edited texts do not represent the exact words of the singer who dictated them. Marjanovic brought to the editing criteria different from those of the singer. Sometimes he omitted 'awkward' lines, such as 'Then you should see Beg Mustajbeg', lines the singer used frequently in performance and, interestingly enough, continued to use in dictating the text for a scribe. (13)

The changes included the removal of verbal repetitions. (14) Marjanovic's procedure reminds me especially of the liberties taken by Zenodotus. (15)

The vestiges of dictation include narrative inconsistencies, which are common. For instance, on the morning when Odysseus is to slay the suitors, Zeus thunders 'from the clouds' as an omen to the hero, who is suitably gladdened:

autika d' ebrontesen ap' aigleentos Olympou,
hypsothen ek nepheon gethese de dios Odysseus,
phemen d' ex okoio gune proeeken aletris
plesion, enth' ara hoi mylai eiato poimeni laon.
(Od. 20.103-6)

Next, an old servant hears the thunder and exclaims to Zeus that 'there is no cloud anywhere':

he rha mulen stesasa epos phato, sema anankti:
Zeu pater, os te theoisi kai anthropoisin anasseis,
e megal' ebrontesas ap' ouranou asteroentos,
outde pothi nephos esti: teras nu teoi tode phaineis
. (Od. 20.113-14)

Then Odysseus rejoices at her lucky utterance and at Zeus' thunder. Atttempts to remove the contradiction by deleting line 10416 are unconvincing, as they do not explain how it arose; this is a case where the poet anticipates himself, the fault most often found in the songs of Avdo Mededovic, as Lord noted. (17) In admitting this contradiction, neither Homer nor his putative editor makes any use of the technology of writing to correct it. How remarkable that it is still in our text!

Without dictation, we would have no text of Homer. But dictation is a necessary, but not a sufficient condition to explain that fact. Any period of 'memorial transmission' before the idea of a fixed text had arisen is not credible in the light of the comparative evidence; the poem would have been substantially altered by such a transmission, (18) or would have developed variant recensions. But we have only one version of the epics, not several, as in the case, for instance, in the Song of Roland or Digenis Akritas. (The recent claim that the Alexandrians knew 131 different versions of the Homeric poems is a mistake, based on T.W. Allen's calculation that the 'city' texts, a group of texts containing scholarly emendations, are mentioned 131 times in the scholia.)

As Lord argued, (19) the impetus to dictation is unlikely to come from an oral-traditional singer, but rather from someone else, a collector or patron. One influence on the person responsible for the recording must have been knowledge of the existence of written literature, which means the written epics of the Levant. (20) That person also knew the alphabet as adapted from Phoenician to Greek. (21) Whoever it was who caused the poems to be written down had considerable time and resources. The dictation by Avdo Mededovic of The Wedding of Smailagic Meho took two entire weeks. (22) In eighth-century Ionia, the number of leather rolls needed for a similar task represented a major expense. This limits us to the courts of wealthy princes and nobles. But all this still does not explain why the recording happened.

Only a cultural or ideological motivation can account for the desire to record these epics. I cannot here enter into the debate on the ideological tendencies of Homer's poems, so you must be content with a brief argument for my own position. It is usually assumed (23) that the epics support an aristocratic ideology. I disagree; it is monarchy that appears most prominently in both epics. In the Iliad Agamemnon misuses his power, and in the Odyssey Odysseus regains his kingdom from a bunch of aristocratic usurpers. I suspect that a king or prince had two such epics recorded in writing because of the ideological support which the epics of this singer--Homer--could offer to traditional images of authority. This was precisely the time--the eighth century--when the weakly-rooted Dark-Age monarchies, seeking legitimacy from oral- traditional memories of, and fictions about, Mycenaean dynasties, were successfully challenged by new aristocratic elites. (24) The written transcripts were kept on Chios among the Homeridae. Rhapsodes performed parts of them from memorization, but the Homeric epics had only a limited popularity until the sixth century, when monarchs like Pisistratus, seeking to buttress their authority against the aristocracy, revived and popularised them. The Pisistratids procured a copy in order to regulate the sequence of rhapsodic performances at Athens. This Athenian set of rolls was the main channel through which the poems were transmitted to Alexandria. (25) Yet we can still identify the text of an archetype, because Parry's words still hold true: 'the seeming vagaries of the manuscript tradition accord with the processes of oral poetry and thus bear witness of their faithfulness'. (26)


(1) M. Parry, The Making of Homeric Verse, ed. A. Parry, Oxford 1971, 451 (written in January 1934).

(2) The Language of Hesiod, Oxford 1972.

(3) For a survey see B.B. Powell, Homer and the Origin of the Greek Alphabet, Cambridge 1991, 186-220. I agree with West, art. cit. 207, that representations of the Iliad do not antedate c. 625; but those of the Odyssey go back to at least c. 660 (Powell, op. cit. 211). The arguments of H. van Wees (Greece &Rome 41 (1994) 1-18, 131-55) neglect the likelihood that vase-painters are representing heroic battles, with a mixture of weaponry characteristic of different dates, and are therefore poor evidence for contemporary warfare. West's recent claim (art. cit. 211-9) that the destruction by flood of the Achaean wall at Il. 12.17-33 is inspired by the destruction of Babylon by Sennacherib in 689 B.C. rests on an archaeological misconception: eighth- century Greeks were perfectly familiar with construction in mud-brick, often on a stone socle; this was normal and need not suggest borrowing from the Near East (p. 213). The effects of torrential rains on unprotected walls of mud-brick are apparent from most excavations of Bronze- and Iron-Age sites in Greece (at Ayios Stephanos in Laconia in my own experience), and would have been easily observable in the eighth century.

(4) 'D'Homère aux origines proto-mycŽniennes de la tradition épique', 1-96 in J.P. Crielaard (ed.), Homeric Questions, Amsterdam 1995, esp. 21-6. For similar datings see A.C. Cassio, 'KEINOS, KALLISTEFANOS e la circolazione dell'epica in area euboica', Annali di Archeologia e Storia Antica 1 (1994) 55-67, esp. 64 n. 66.

(5) Oral Poetry, Cambridge 1977, 18-22.

(6) For this misunderstanding cf. e.g. Taplin, op. cit. 36.

(7) A.B. Lord, The Singer of Tales, Cambridge MA 1960, 21.

(8) Ibid. 28; thus Thomas, op. cit. 36-8, is wrong to imply that this is not Lord's view.

(9) For this error see Thomas, loc. cit.

(10) See Lord, op. cit. (n. 11), 11, 20, 197-200.

(11) Ibid. 181.

(12) Op. cit. (n. 24), 128.

(13) Lord, op. cit. (n. 11), 16. For other instances cf. J.M. Foley, Traditional Oral Epic, Berkeley and Los Angeles 1990, 28 n. 18.

(14) Ibid. 17-18.

(15) Cf. The Iliad: A Commentary. IV: Books 13-16, Cambridge 1992, 22-5. Another tendency in the MSS of Homer is the standardisation of repeated verses against each other: see my article 'The Iliad and its editors: Dictation and Redaction', Classical Antiquity 9 (1990) 326-34, esp. 332-3 (published in Italian translation as 'L'Iliade fra dettatura e redazione', SIFC 10 (1992) 833-43, esp. 840-2).

(16) E.g. R.B. Rutherford, Homer, Odyssey Books XIX and XX, Cambridge 1992, ad loc.

(17) Op. cit. (n. 8), xii.

(18) So the theory of G. Nagy, most clearly expressed in his article cited in n. 7, that the progressively wider and wider diffusion of the Homeric poems resulted in their gradually becoming more and more fixed. However, the reverse outcome would seem more likely, as is indeed supported by the plethora of early papyrus texts with inorganic additional lines; and one is entitled to ask why the resulting texts contain so many minor oddities, which would surely have been tidied up in any process of this kind. Also, my linguistic researches (n. 2, and The Iliad: A Commentary. IV: Books 13-16, 14 with n. 19, 17 with n. 28) have shown that the texts were fixed at different linguistic stages--a most unlikely outcome for any process other than either dictation or fixation in writing! Nagy's theory is followed by Foley, op. cit. (n. 49) 21-31; he falls into the trap of thinking that the 'wild' papyri are different versions rather than merely replete with inorganic plus-verses (26). He also misses the fact that other poems in the tradition survive (15), and claims that the Alexandrians knew 131 separate editions of Homer (24, 28)!

(19) Art. cit. (n. 7), = op. cit. (n. 9), esp. 44.

(20) On the probable influence of the latter on Homer cf. W. Burkert, The Orientalizing Revolution, trans. M.E. Pinder & W. Burkert, Cambridge MA 1992, 114-20.

(21) For the theory that the same person who adapted the alphabet did so in order to take down dictation the Homeric poems, cf. Powell, op. cit. (n. 5).

(22) A.B. Lord, 'Homer, Parry and Huso', American Journal of Archaeology 52 (1948) 34-44, esp. 42 (reprinted in M. Parry op. cit. (n. 1), 476; cf. Lord op. cit. (n. 8), 12.

(23) E.g. Ian Morris, 'The Use and Abuse of Homer', Classical Antiquity 5 (1986) 81-138.

(24) On early kingship see P. Carlier, La Royauté en Grèce avant Alexandre, Paris 1984, esp. 195-214 on Homer; J. Lenz, Kings and the Ideology of Kingship in Early Greece, Diss. Columbia 1993.

(25) For a fuller account of my view of the transmission cf. The Iliad: A Commentary. IV: Books 13-16, 20-37.

(26) Parry op. cit. (n. 1), 353 n. 1.

Richard Janko
University College, London
E-mail: R.Janko@ucl.ac.uk

(Richard Janko is Professor of Greek at University College London.)